Are you really on the map?

Between 18^th July and 18^th September 2012, Research Office heads or the PVCs Research (or their equivalent) in one hundred UK institutions in receipt of EPSRC funding, were invited to complete an online DCC survey designed to establish reactions to (and the impact from) EPSRC’s call for the development of effective research data roadmaps. For the purpose of this survey we identified HEIs that receive more than 10% of their total funding from research and are currently in receipt of one or more EPSRC grants. This gave us our list of one hundred institutions. Of that one hundred, all of whom were invited to take part, thirty-five completed the survey.

Whilst the EPSRC is a significant funder for many of the one hundred institutions contacted, almost half have fewer than ten EPSRC grants each, with a combined value of £49.5 million. Another thirty-seven institutions have between ten and ninety-nine grants each, with a total value of £1.07 billion, while the fourteen institutions that each have more than one hundred grants currently account for £2.4 billion in EPSRC funding.

Of the forty-nine institutions with fewer than ten grants, twenty-two responded to our survey. There were also responses from eleven institutions with between ten and ninety-nine grants; but none were forthcoming from the ten institutions with between one hundred and two hundred grants. Of the four that hold more than two hundred grants, two did respond. Such a bias in the categories of institution responding makes it difficult to draw strong conclusions from the survey, although one may observe that participation by the lower end forty-nine, which produced the better response rate, tended to be from the more modern and often smaller institutions. One may conjecture further that in these institutions it has been easier to assemble a representative view of the EPSRC expectations and roadmap.

So what did we want to know? Question 1 of our survey asked respondents to select from statements that might describe the status of their institution's progress towards meeting EPSRC expectations regarding its Policy Framework on Research Data. Reassuringly, none chose the option ‘we are not aware that this policy applies to our institution’, although there were three that admitted to being in the very early stages, having acknowledged the EPSRC requirements for a roadmap but were still at the point of considering a response. More positively, thirteen had identified a working group to formulate a roadmap, ten had produced a roadmap for endorsement by senior management and nine declared they have roadmaps that have been endorsed by senior management. Overall, it is evident that while the need to develop roadmaps has been embraced, the May 2012 deadline has not been met by 45% of the responding institutions, which inevitably leaves them less time in which to implement their roadmaps. We have no hard evidence to show whether this reflects the condition amongst the non-respondent institutions, although anecdotal evidence suggests that such a situation is possible.

In a supplement to Question 1 we also asked institutions that had completed a roadmap to indicate what had been included in the scope for them, which produced the following results:

Additionally, one institution referred to its roadmap as a very high-level description of the pathway including rough milestones to May 2015, with another describing its roadmap in terms of an annex to a discussion paper that sets out the issues the University needs to address in order to achieve compliance with EPSRC requirements – which sounds like a draft roadmap. This latter institution is also currently recruiting a Research Data Programme Manager in order to progress the data management agenda, which further underwrites their commitment to deliver.

The majority (eighteen) of respondents have included in their roadmap a list of tasks and a timetable in which to complete them; but only nine have assigned roles and responsibilities, and a mere eight have fully described the resources required, which does not augur well for completion of the three year implementation programme. It is of particular concern that where roadmaps have already been endorsed by senior management only three have assigned roles and responsibilities or give a description of the resources required to achieve sustainable implementation.

The DCC has consistently advocated that the development of strategies and policies for research data management should actively be championed by senior managers and that working groups and consultative committees must be inclusive when dealing with the research community. We were therefore encouraged by the answers given to the question ‘which of the following stakeholder groups have been included when making decisions about an EPSRC roadmap?’:

In addition to the groups in our table, other stakeholders were identified by four respondents, comprising the Archives, the Governance Office, the Research Ethics Committee and ‘other HEIs’. Most institutions responding to the survey are therefore utilising the skills and knowledge available to them from across their communities, which should lead to the early identification of the skills they already have and the skills gaps they will need to fill. That said, it is disappointing to note that not all thirty-five institutions have involved researchers in roadmap design, since without their ‘buy-in’ it will be more difficult to implement new systems and services that are appropriate, desirable and shaped in a manner that will not introduce new barriers to the research process.

For many institutions, having to develop a roadmap when previously there had been little corporate engagement with the intricacies of research data management, was always likely to present fresh logistical and technical challenges, and we were keen to ask what they regarded as the main obstacles to compliance with the EPSRC’s expectations. The results, in a table where as many options could be selected as deemed appropriate, were as follows:

Further comments were offered by eight respondents, one of whom said there were no particular obstacles and another that the institution-wide Research Data Management Project is already addressing skills, processes and techniques. The others contended

• the challenge of designing appropriate policies to meet ethical considerations, cross-disciplinary research and multi-institutional projects, especially where international collaboration might be in evidence;
• a lack of clarity for some key concepts (e.g. what exactly constitutes a data set?);
• the possibility that a 'least effort required' option may be preferred by those making the final decision, rather than proactively seeking to meet best practice;
• the scale and sheer diversity of research and data types, the volume of data, PIs questioning the demand for 'public access', questions over what can be done sector wide rather than HEI by HEI, the lack of precedents or exemplars and technical issues related to curation;
• the risk that creating sound business and sustainability models could lead to a lack of forward capacity;
• the difficulty for a small research office of pulling the roadmap together and achieving buy-in from senior management.

Not unexpectedly, the main obstacles identified by respondents were the availability of appropriate human and technological infrastructure, funding and skills. Research data management has, on the whole, been the realm of particular disciplinary groups or individual researchers and, apart from the provision of file space for data storage, it is generally not an area to which significant consideration has been devoted at an institutional level. The view that low management priority persists in seven of the responding institutions is especially unfortunate when the main research funders have for some years now been explicit about the need for effective data management processes to be embedded in institutional business. Whilst the institutions to have identified low management priority are those with the smallest number of EPSRC grants, which may suggest that data management features less prominently in the broader business portfolio, they are no less subject to risks from poor management processes or the failure to comply with data legislation that have dogged some of our larger institutions in recent years.

On the flip-side to perceived obstacles, we also asked how helpful are the EPSRC’s nine expectations to the organisation and management of research data in general. One enthusiastic respondent welcomed the EPSRC approach as extremely useful in galvanising institutional action and, judged by most to be broadly in line with RCUK’s common principles on data policy, twenty-four respondents agreed they are appropriate to data generated from research funded by other agencies, with twenty taking them as a firm foundation for the development of good practice to meet a broad range of expectations. This is a good proportion but, given the closeness of the EPSRC’s Policy Framework to the RCUK’s Principles, perhaps one might have expected something closer to 100%. Further investigation of this aspect of the survey might be warranted. Nonetheless, only a single respondent went so far as to declare the expectations relevant solely to EPSRC-funded research data. For others, their generally positive attitude was tempered by a view of the expectations as reasonable but exceptionally difficult to meet, the range of perspectives itself compounding the lack of consensus, with researchers in general seen as regarding the expectations as another hurdle to jump through, despite agreeing with the underlying concepts; administrators recognising the need but complaining about adding another element to the already lengthy and cumbersome pre-award submission process; and ethics committees concerned over potential conflicts with ethical concerns such as the risk of inadvertent disclosure. Some concerns emanated from the lack of a clear understanding of what is expected, one view expressed being that the EPSRC's expectations come from an predominantly administrative perspective (i.e. access to data to support published research rather than all data) and three reported a lack of clarity of detail.

Notwithstanding these caveats, if this limited sample is truly indicative of a more general view, it suggests that most institutions now understand (or believe they understand) what is expected of them. More marked was the over-riding ambivalence of opinion reflected by the gulf between the group of sixteen who saw the expectations as an opportunity to focus attention and resources on a neglected aspect of research and the fifteen respondents who subscribed to the opinion that the expectations are difficult to reconcile with prevailing financial conditions, referring directly to current budgetary constraints.

So we are left with a message that, all in all, at least as far as our sample cohort is concerned, the nine EPSRC expectations are a good thing if one can only afford to take them on board. This takes us back to the subject of management commitment and the prioritisation of institutional resources.

Whether that view is shared across the full suite of one hundred EPSRC grant recipients remains unknown. But if one can speculate that their failure to respond to the survey was at least in part due to the complexity of their individual arrangements, the odds against it being the case are certainly shortened.

Turning to future requirements for support, we were keen to understand what kinds of assistance institutions will need when designing or implementing data roadmaps and policies, in order to inform DCC forward planning. Seven respondents replied with the message that no support needs were currently predicted but, overall, the comments made confirm that the respondent institutions in general are as yet in the very early stages of coming to terms with what is implied by effective research data management. However, whilst the majority continues to demand broad guidance and generic frameworks that will help them define their research data management programmes, the responses made show the beginnings of a demand for more explicit help, for example in the selection of standards and protocols, for hands-on training in data management techniques, in the provision of independent progress monitoring and assistance in process modelling. For the DCC, the key message is that institutions are likely to be seeking increasingly focused interventions as they embark on their implementation programmes, which demand is likely to accelerate now that the clock is well and truly counting down the hours to May 2015.